bin/mog

   1 #!/usr/bin/env ruby
   2 require 'mogilefs'
   3 require 'optparse'
   4 $stdin.binmode
   5 $stdout.binmode
   6 $stderr.sync = $stdout.sync = true
   7
   8 trap('INT') { exit 130 }
   9 trap('PIPE') { exit 0 }
  10
  11 # this is to be compatible with config files used by the Perl tools
  12 def parse_config_file!(path, overwrite = false)
  13   dest = {}
  14   File.open(path).each_line do |line|
  15     line.strip!
  16     if /^(domain|class)\s*=\s*(\S+)/.match(line)
  17       dest[$1.to_sym] = $2
  18     elsif m = /^(?:trackers|hosts)\s*=\s*(.*)/.match(line)
  19       dest[:hosts] = $1.split(/\s*,\s*/)
  20     elsif m = /^timeout\s*=\s*(.*)/.match(line)
  21       dest[:timeout] = m[1].to_f
  22     else
  23       warn "Ignored configuration line: #{line}" unless /^#/.match(line)
  24     end
  25   end
  26   dest
  27 end
  28
  29 # parse the default config file if one exists
  30 def_file = File.expand_path("~/.mogilefs-client.conf")
  31 def_cfg = File.exist?(def_file) ? parse_config_file!(def_file) : {}
  32
  33 # parse the command-line first, these options take precedence over all else
  34 cli_cfg = {}
  35 config_file = nil
  36 ls_l = false
  37 ls_h = false
  38 test = {}
  39 cat = { :raw => false }
  40
  41 ARGV.options do |x|
  42   x.banner = "Usage: #{$0} [options] <command> [<arguments>]"
  43   x.separator ''
  44
  45   x.on('-c', '--config=/path/to/config',
  46        'config file to load') { |file| config_file = file }
  47
  48   x.on('-t', '--trackers=host1[,host2]', '--hosts=host1[,host2]', Array,
  49        'hostnames/IP addresses of trackers') do |trackers|
  50     cli_cfg[:hosts] = trackers
  51   end
  52
  53   x.on('-e', 'True if key exists') { test[:e] = true }
  54   x.on('-r', '--raw', 'show raw big_info file information') { cat[:raw] = true }
  55
  56   x.on('-C', '--class=s', 'class') { |klass| cli_cfg[:class] = klass }
  57   x.on('-d', '--domain=s', 'domain') { |domain| cli_cfg[:domain] = domain }
  58   x.on('-l', "long listing format (`ls' command)") { ls_l = true }
  59   x.on('-h', '--human-readable',
  60        "print sizes in human-readable format (`ls' command)") { ls_h = true }
  61
  62   x.separator ''
  63   x.on('--help', 'Show this help message.') { puts x; exit }
  64   x.parse!
  65 end
  66
  67 # parse the config file specified at the command-line
  68 file_cfg = config_file ? parse_config_file!(config_file, true) : {}
  69
  70 # read environment variables, too.  This Ruby API favors the term
  71 # "hosts", however upstream MogileFS teminology favors "trackers" instead.
  72 # Favor the term more consistent with what the MogileFS inventors used.
  73 env_cfg = {}
  74 if ENV["MOG_TRACKERS"]
  75   env_cfg[:hosts] = ENV["MOG_TRACKERS"].split(/\s*,\s*/)
  76 end
  77 if ENV["MOG_HOSTS"] && (env_cfg[:hosts] || []).empty?
  78   env_cfg[:hosts] = ENV["MOG_HOSTS"].split(/\s*,\s*/)
  79 end
  80 env_cfg[:domain] = ENV["MOG_DOMAIN"] if ENV["MOG_DOMAIN"]
  81 env_cfg[:class] = ENV["MOG_CLASS"] if ENV["MOG_CLASS"]
  82
  83 # merge the configs, favoring them in order specified:
  84 cfg = {}.merge(def_cfg).merge(env_cfg).merge(file_cfg).merge(cli_cfg)
  85
  86 # error-checking
  87 err = []
  88 err << "trackers must be specified" if cfg[:hosts].nil? || cfg[:hosts].empty?
  89 err << "domain must be specified" unless cfg[:domain]
  90 if err.any?
  91   warn "Errors:\n  #{err.join("\n  ")}"
  92   warn ARGV.options
  93   exit 1
  94 end
  95
  96 unless cmd = ARGV.shift
  97   warn ARGV.options
  98   exit 1
  99 end
 100
 101 cfg[:timeout] ||= 30 # longer timeout for interactive use
 102 mg = MogileFS::MogileFS.new(cfg)
 103
 104 def store_file_retry(mg, key, storage_class, filepath)
 105   tries = 0
 106   begin
 107     mg.store_file(key, storage_class, filepath)
 108   rescue MogileFS::UnreadableSocketError,
 109          MogileFS::Backend::NoDevicesError => err
 110     if ((tries += 1) < 10)
 111       warn "Retrying on error: #{err}: #{err.message} tries: #{tries}"
 112       retry
 113     else
 114       warn "FATAL: #{err}: #{err.message} tries: #{tries}"
 115     end
 116     exit 1
 117   end
 118 end
 119
 120 def human_size(size)
 121   suff = ''
 122   %w(K M G).each do |s|
 123     size /= 1024.0
 124     if size <= 1024
 125       suff = s
 126       break
 127     end
 128   end
 129   sprintf("%.1f%s", size, suff)
 130 end
 131
 132 begin
 133   case cmd
 134   when 'cp'
 135     filename = ARGV.shift or raise ArgumentError, '<filename> <key>'
 136     key = ARGV.shift or raise ArgumentError, '<filename> <key>'
 137     ARGV.shift and raise ArgumentError, '<filename> <key>'
 138     store_file_retry(mg, key, cfg[:class], filename)
 139   when 'cat'
 140     ARGV.empty? and raise ArgumentError, '<key1> [<key2> ...]'
 141     ARGV.each do |key|
 142       if (!cat[:raw] && key =~ /^_big_info:/)
 143         mg.bigfile_write(key, $stdout, {:verify => true})
 144       else
 145         mg.get_file_data(key, $stdout)
 146       end
 147     end
 148   when 'ls'
 149     prefixes = ARGV.empty? ? [ nil ] : ARGV
 150     if ls_l
 151       each_key = lambda do |key, size, devcount|
 152         size = ls_h && size > 1024 ? human_size(size) : size.to_s
 153         size = (' ' * (12 - size.length)) << size # right justify
 154         puts [ sprintf("% 2d", devcount), size, key ].pack("A4 A16 A*")
 155       end
 156     else
 157       each_key = lambda { |key| puts key }
 158     end
 159     prefixes.each { |prefix| mg.each_key(prefix, &each_key) }
 160   when 'rm'
 161     ARGV.empty? and raise ArgumentError, '<key1> [<key2>]'
 162     ARGV.each { |key| mg.delete(key) }
 163   when 'mv'
 164     from = ARGV.shift or raise ArgumentError, '<from> <to>'
 165     to = ARGV.shift or raise ArgumentError, '<from> <to>'
 166     ARGV.shift and raise ArgumentError, '<from> <to>'
 167     mg.rename(from, to)
 168   when 'stat' # this outputs a RFC822-like format
 169     ARGV.empty? and raise ArgumentError, '<key1> [<key2>]'
 170     ARGV.each_with_index do |key, i|
 171       if size = mg.size(key)
 172         puts "Key: #{key}"
 173         puts "Size: #{size}"
 174         mg.get_paths(key).each_with_index do |path,i|
 175           puts "URL-#{i}: #{path}"
 176         end
 177         puts ""
 178       else
 179         warn "No such key: #{key}"
 180       end
 181     end
 182   when 'tee'
 183     require 'tempfile'
 184     key = ARGV.shift or raise ArgumentError, '<key>'
 185     ARGV.shift and raise ArgumentError, '<key>'
 186     cfg[:class] or raise ArgumentError, 'E: --class must be specified'
 187     buf = ''
 188     tmp = Tempfile.new('mog-tee') # TODO: explore Transfer-Encoding:chunked :)
 189
 190     # if stdout is pointing to /dev/null, don't bother installing the filter.
 191     tee_obj = tmp
 192     if File.stat('/dev/null') != $stdout.stat
 193       tee_obj = lambda do |buf|
 194         rv = nil
 195         [ $stdout, tmp ].each { |io| rv = io.write(buf) }
 196         rv
 197       end
 198       def tee_obj.write(buf)
 199         self[buf]
 200       end
 201     end
 202     begin
 203       MogileFS::X.copy_stream($stdin, tee_obj)
 204       store_file_retry(mg, key, cfg[:class], tmp.path)
 205     ensure
 206       tmp.close!
 207     end
 208   when 'test'
 209     truth, ok = true, nil
 210     raise ArgumentError, "-e must be specified" unless (test.size == 1)
 211
 212     truth, key = case ARGV.size
 213     when 1
 214       [ true, ARGV[0] ]
 215     when 2
 216       if ARGV[0] != "!"
 217         raise ArgumentError, "#{ARGV[0]}: binary operator expected"
 218       end
 219       [ false, ARGV[1] ]
 220     else
 221       raise ArgumentError, "Too many arguments"
 222     end
 223
 224     paths = mg.get_paths(key)
 225     if test[:e]
 226       ok = !!(paths && paths.size > 0)
 227     else
 228       raise ArgumentError, "Unknown flag: -#{test.keys.first}"
 229     end
 230
 231     truth or ok = ! ok
 232     exit ok ? 0 : 1
 233   else
 234     raise ArgumentError, "Unknown command: #{cmd}"
 235   end
 236 rescue ArgumentError => err
 237   warn "Usage: #{$0} #{cmd} #{err.message}"
 238   exit 1
 239 end
 240 exit 0